PS3 or gaming computer?

Cheers for the advice, think Im gonna get the PS3 with the new GTA. Its 439euro with the game. Rang a shop today and apprently there opening at 12am tonight for the GTAs release!!!! ha ha thats fecking mental
 
Decotron said:
Cheers for the advice, think Im gonna get the PS3 with the new GTA. Its 439euro with the game. Rang a shop today and apprently there opening at 12am tonight for the GTAs release!!!! ha ha thats fecking mental
Click to expand...

they do that alot here
for almost every big game opening
 
Redlambs said:
to get similar results on the PS3 architecture would be an interesing job.

I must say, it's interesting to see what they do though, it is really intriguing me.
Click to expand...

PS3 in the end is limited by bandwidth and RAM, as there is a finite number of pixels that the 500Mhz RSX can push using its 8 ROPs on a 128bit bus to 256MB of GDDR3 and a finite amount of space to push them to. Cell is no rasterizer, RSX is in a different league in this respect, but the Cell can be used to do a lot of tasks early on in the graphics pipeline and a lot of post processing at the end of it in addition to the more general game processing work. Developers however need to redesign their render pipelines for PS3 for this to be of any significance, whereas on the PC and XB360 the only real choice is to give everything to the GPUs.

PS3 Cell SPUs are not specific to graphical tasks as GPU shading unit are, but then, they are far more flexible, run one hell of a lot faster, and each have their own local memory. They can act as individual parallel processors DMAing data in, or true stream processors with one passing its result data to the next over the internal bus for more processing, including then directly onto the RSX. This can obviously be arranged in quite a decent number of configurations.

It will not be known how the PS3 really performs until a third party developer does something really well on the PS3 first, and then tries to get it working on the XB360 and the PC. Only when this happens will we see how the latter's brute force GPU only approach compares (a high end PC would win the competition now, in regards to the PC comparison total performance is not the question, but rather bang for buck). There are quite a few examples where the PS3 performs poorly in comparison when the process is directly applied in the opposite direction and is used in this GPU only mode, with the classic one probably being Valve's The Orange Box. It is known that TOB on the PS3 does not use any of the SPUs at all, none, they sit there and twiddle their thumbs while the single core PPE does all of the game work and the RSX all the graphics - where the sound is processed I do not know, but it is not an SPU. This is a game with code that I suppose is originally written and optimized to run on a Core 2 Duo and a Shader Model 3 class GPU. It is no surprise that it does not end well when run on the PS3. Some developers state that you can still use bad programming practices and data design and just about get away with it on the multi-core environment of the XB360, but it you will drown doing this on PS3.

What is interesting here is when developers complain that it is too hard to get the same results out of the PS3. On the PC and XB360, they can use their tools used their mature toolsets, tell the API what to do, and the hardware readily goes about the job and does it with no fuss or messing. For the PS3, they have to then mess around like hell just to get the product even comparable. The problem here is that at best that is all you are going to get, something comparable, but you will still not be using the PS3 hardware to anywhere near its peak performance.

It is a lot of little things, right. I mean, we had an idea of what to expect with the Cell before, but there were definitely some hard-learnt lessons. One that I can think of is to not treat the SPU as a co-processor. It is not a way to take work off of the PPU, that is a wrong-headed view of the SPU – it is all about programming for the SPUs first and putting stuff on the PPU only when necessary. You don't use the SPUs to optimise the PPU.
Click to expand...

Remember that SPU local memory is really fast: Classically, you want to avoid algorithms that have heavy memory access. That's still true in general, just not on the SPUs themselves.
Click to expand...

Now, considering that an SPU currently only has 256K of local memory for both code and data, big gloaty data structures and code branching all over the show designed for other systems is not going to fit well. If these things are not designed specifically for the SPU, then again, it is not going to work out the best. This is just talking about getting the game to run well on the Cell and use the RSX as the GPU to get reasonable performance, without looking at how you can use the SPUs to do graphical work and other things that they are theoretically very good at to increase system performance further. This is without looking at how to use the distributed non unified memory model and the bandwidth advantages it can bring to the table. This is the problem, and I personally cannot see a lot of third party developers making the effort to design their games from scratch using methodologies applicable to the Cell programming model, even though it seems likely that doing so will also bring advantages to the other platforms as well (yes XBots, developing for the PS3 first will get better performance out of the Microsoft machine).

The Cell is a paradox to likes of developers such as Valve, yet the true paradox is their continued aversion to Cell like architectures and seeming unwillingness to embrace its philosophy when it is quite clear that the hardware industry as a whole, not just consoles or gaming, is moving with great speed towards a Cell-like future. The barriers of physics finally gave them a slap. Intel are increasing the number of cores in their legacy processors and investing significant amounts in the Larabee project, the likes of nVidia and AMD are pushing the boundaries of their GPU processing paradigm further and further towards more general processing tasks, while IBM is working on Cell 2. All of this is going to meet in the middle somewhere, and the world will have changed leaving the likes of Valve in a deep muddy hole. The developers that embrace the Cell now will not only get good performance out of their non-Cell targets, but will be perfectly placed when the next phase of gaming hardware comes along. This is to some extent speculation however. ;)
 
In the end it comes down to what games you play.
Most MMO:s and RTS games will never be released on a console or at least very few.
However you want to play awesome games like halo (sarcasm!!!) then get a PS3/Xbox360. But for the love of good dont buy a PC and get ported console games in 9 out of 10 cases things go bad.
 
good. fwah you had me thinking there that you spent a good 2 hours on writing that. But it was just a quick cut and paste job. your a proper man.
 
You must log in or register to reply here.